Re: is there a fix for ubuntu with intel GPU booting to blank screen?

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Anca Emanuel]

On Mon, Dec 27, 2010 at 1:50 PM, Robert P. J. Day <rpjday@crashcourse.ca> wrote:
>
>  i've posted on a couple other lists but i figured i'd ask the
> experts.  on my updated/upgraded ubuntu system (gateway NV79 with
> intel GPU), i've built my own kernel from the latest linus git tree as
> i've done many times in the past, but this kernel boots to a blank
> screen.  here's a reference that seems to nicely summarize my
> situation:
>
>  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/561802
>
>  weirdly, i'm currently running an earlier "rc" hand-rolled git
> kernel which works fine:
>
> $ uname -a
> Linux lynx 2.6.37-rc5+ #1 SMP Fri Dec 10 12:09:02 EST 2010 x86_64 GNU/Linux
> $
>
> (even though it says lynx, it's maverick simply fully updated.)
>
>  and this is from lspci:
>
> 00:02.0 VGA compatible controller: Intel Corporation Core Processor
> Integrated Graphics Controller (rev 12) (prog-if 00 [VGA controller])
>        Subsystem: Acer Incorporated [ALI] Device 031c
>        Flags: bus master, fast devsel, latency 0, IRQ 41
>        Memory at d0000000 (64-bit, non-prefetchable) [size=4M]
>        Memory at c0000000 (64-bit, prefetchable) [size=256M]
>        I/O ports at 3050 [size=8]
>        Expansion ROM at <unassigned> [disabled]
>        Capabilities: <access denied>
>        Kernel driver in use: i915
>        Kernel modules: i915
> ... snip ..
>
>  i've tried various suggestions from google, as in "modeset noacpi"
> and others, but nothing works to get video on my rc7+ kernel.  the
> rc5+ verison, though, boots properly without incident.  and the link i
> gave to launchpad above suggests, at the bottom, that even as late as
> dec 12, people were still having problems.
>
>  i'm open to suggestions, not scared to patch my kernel source and
> rebuild and reboot.
>
> rday
>
> --
>
> ========================================================================
> Robert P. J. Day                               Waterloo, Ontario, CANADA
>                        http://crashcourse.ca
>
> Twitter:                                       http://twitter.com/rpjday
> LinkedIn:                               http://ca.linkedin.com/in/rpjday
> ========================================================================
>
> --
> kernel-team mailing list
> kernel-team@lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>

CC: lkml

did you tried 2.6.37-rc8-git1 ?

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Robert P. J. Day]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 3961 bytes --]

On Fri, 31 Dec 2010, Anca Emanuel wrote:

> On Mon, Dec 27, 2010 at 1:50 PM, Robert P. J. Day <rpjday@crashcourse.ca> wrote:
> >
> >  i've posted on a couple other lists but i figured i'd ask the
> > experts.  on my updated/upgraded ubuntu system (gateway NV79 with
> > intel GPU), i've built my own kernel from the latest linus git tree as
> > i've done many times in the past, but this kernel boots to a blank
> > screen.  here's a reference that seems to nicely summarize my
> > situation:
> >
> >  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/561802
> >
> >  weirdly, i'm currently running an earlier "rc" hand-rolled git
> > kernel which works fine:
> >
> > $ uname -a
> > Linux lynx 2.6.37-rc5+ #1 SMP Fri Dec 10 12:09:02 EST 2010 x86_64 GNU/Linux
> > $
> >
> > (even though it says lynx, it's maverick simply fully updated.)
> >
> >  and this is from lspci:
> >
> > 00:02.0 VGA compatible controller: Intel Corporation Core Processor
> > Integrated Graphics Controller (rev 12) (prog-if 00 [VGA controller])
> >        Subsystem: Acer Incorporated [ALI] Device 031c
> >        Flags: bus master, fast devsel, latency 0, IRQ 41
> >        Memory at d0000000 (64-bit, non-prefetchable) [size=4M]
> >        Memory at c0000000 (64-bit, prefetchable) [size=256M]
> >        I/O ports at 3050 [size=8]
> >        Expansion ROM at <unassigned> [disabled]
> >        Capabilities: <access denied>
> >        Kernel driver in use: i915
> >        Kernel modules: i915
> > ... snip ..
> >
> >  i've tried various suggestions from google, as in "modeset noacpi"
> > and others, but nothing works to get video on my rc7+ kernel.  the
> > rc5+ verison, though, boots properly without incident.  and the link i
> > gave to launchpad above suggests, at the bottom, that even as late as
> > dec 12, people were still having problems.
> >
> >  i'm open to suggestions, not scared to patch my kernel source and
> > rebuild and reboot.
>
> CC: lkml
>
> did you tried 2.6.37-rc8-git1 ?

  i just did a "git pull" to get the latest (-rc8) and "git log" shows
me:

commit 3643e0e87c13c670a0fdcd0c34401b38b36ff021
Merge: e983dc2 cc6455f
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu Dec 30 12:09:26 2010 -0800

    Merge branch 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ickle/drm-intel

    * 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ickle/drm-intel:
      drm/i915/dvo: Report LVDS attached to ch701x as connected
      Revert "drm/i915/bios: Reverse order of 100/120 Mhz SSC clocks"
      drm/i915: Verify Ironlake eDP presence on DP_A using the capability fuse
      drm/i915, intel_ips: When i915 loads after IPS, make IPS relink to i915.
      drm/i915/sdvo: Add hdmi connector properties after initing the connector
      drm/i915: Set the required VFMUNIT clock gating disable on Ironlake.


  no change, still prints "failed to load i915 symbols, graphics turbo
disabled" just before it all goes black but it clearly continues to
boot.

  not to sound self-important but this is kind of a show-stopper, is
it not?  right now, i have exactly one bootable kernel (hand-rolled
2.6.37-rc5+) and i have no idea *why* that one boots properly but,
under the circumstances, i can't move on.  i'm willing to do any
testing if that would help the cause but i suspect enough people are
already digging into this so i'll just be patient and test the next
pass to see if that solves it.

rday

-- 

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Maxim Levitsky]

On Fri, 2010-12-31 at 09:23 -0500, Robert P. J. Day wrote:
> On Fri, 31 Dec 2010, Anca Emanuel wrote:
> 
> > On Mon, Dec 27, 2010 at 1:50 PM, Robert P. J. Day <rpjday@crashcourse.ca> wrote:
> > >
> > >  i've posted on a couple other lists but i figured i'd ask the
> > > experts.  on my updated/upgraded ubuntu system (gateway NV79 with
> > > intel GPU), i've built my own kernel from the latest linus git tree as
> > > i've done many times in the past, but this kernel boots to a blank
> > > screen.  here's a reference that seems to nicely summarize my
> > > situation:
> > >
> > >  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/561802
> > >
> > >  weirdly, i'm currently running an earlier "rc" hand-rolled git
> > > kernel which works fine:
> > >
> > > $ uname -a
> > > Linux lynx 2.6.37-rc5+ #1 SMP Fri Dec 10 12:09:02 EST 2010 x86_64 GNU/Linux
> > > $
> > >
> > > (even though it says lynx, it's maverick simply fully updated.)
> > >
> > >  and this is from lspci:
> > >
> > > 00:02.0 VGA compatible controller: Intel Corporation Core Processor
> > > Integrated Graphics Controller (rev 12) (prog-if 00 [VGA controller])
> > >        Subsystem: Acer Incorporated [ALI] Device 031c
> > >        Flags: bus master, fast devsel, latency 0, IRQ 41
> > >        Memory at d0000000 (64-bit, non-prefetchable) [size=4M]
> > >        Memory at c0000000 (64-bit, prefetchable) [size=256M]
> > >        I/O ports at 3050 [size=8]
> > >        Expansion ROM at <unassigned> [disabled]
> > >        Capabilities: <access denied>
> > >        Kernel driver in use: i915
> > >        Kernel modules: i915
> > > ... snip ..
> > >
> > >  i've tried various suggestions from google, as in "modeset noacpi"
> > > and others, but nothing works to get video on my rc7+ kernel.  the
> > > rc5+ verison, though, boots properly without incident.  and the link i
> > > gave to launchpad above suggests, at the bottom, that even as late as
> > > dec 12, people were still having problems.
> > >
> > >  i'm open to suggestions, not scared to patch my kernel source and
> > > rebuild and reboot.
> >
> > CC: lkml
> >
> > did you tried 2.6.37-rc8-git1 ?
> 
>   i just did a "git pull" to get the latest (-rc8) and "git log" shows
> me:
> 
> commit 3643e0e87c13c670a0fdcd0c34401b38b36ff021
> Merge: e983dc2 cc6455f
> Author: Linus Torvalds <torvalds@linux-foundation.org>
> Date:   Thu Dec 30 12:09:26 2010 -0800
> 
>     Merge branch 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ickle/drm-intel
> 
>     * 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ickle/drm-intel:
>       drm/i915/dvo: Report LVDS attached to ch701x as connected
>       Revert "drm/i915/bios: Reverse order of 100/120 Mhz SSC clocks"
>       drm/i915: Verify Ironlake eDP presence on DP_A using the capability fuse
>       drm/i915, intel_ips: When i915 loads after IPS, make IPS relink to i915.
>       drm/i915/sdvo: Add hdmi connector properties after initing the connector
>       drm/i915: Set the required VFMUNIT clock gating disable on Ironlake.
> 
> 
>   no change, still prints "failed to load i915 symbols, graphics turbo
> disabled" just before it all goes black but it clearly continues to
> boot.
> 
>   not to sound self-important but this is kind of a show-stopper, is
> it not?  right now, i have exactly one bootable kernel (hand-rolled
> 2.6.37-rc5+) and i have no idea *why* that one boots properly but,
> under the circumstances, i can't move on.  i'm willing to do any
> testing if that would help the cause but i suspect enough people are
> already digging into this so i'll just be patient and test the next
> pass to see if that solves it.
Since you know how to compile kernel, and even have a close working
version, you really should just bisect the bug.

Bisection is a built-in in the git binary search that allows you to find
the first commit that introduced the bug.

For example, quick google gives,
http://www.kernel.org/pub/software/scm/git-core/docs/v1.0.13/howto/isolate-bugs-with-bisect.txt

Best regards,
	Maxim Levitsky

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Robert P. J. Day]

On Fri, 31 Dec 2010, Maxim Levitsky wrote:

... snip related to i915 blank screen issues ...

> Since you know how to compile kernel, and even have a close working
> version, you really should just bisect the bug.
>
> Bisection is a built-in in the git binary search that allows you to
> find the first commit that introduced the bug.
>
> For example, quick google gives,
> http://www.kernel.org/pub/software/scm/git-core/docs/v1.0.13/howto/isolate-bugs-with-bisect.txt

  oh, i know how to do bisection but, to be frank, i just don't have
the time.  i'm under a brutal deadline for some billable work and,
even if i had the time, i suspect that when i eventually tracked down
the commit, i probably wouldn't really appreciate what i was looking
at, anyway.

  i've done some searching on the issue and, among other pages, i've
come up with this one:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/683775

so i downloaded the kernel mentioned there and tested that, but no
luck.  mostly, i understand that there top men -- *top* *men* --
working on this :-) who know far more about the graphics subsystem
than i do so i'm just going to stay out of their way, but i'm willing
to test any proposed solutions if they come up.  my biggest worry is
that, given that we're already at -rc8, will this issue be fixed
before the next official release?

rday

-- 

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Maxim Levitsky]

On Fri, 2010-12-31 at 12:01 -0500, Robert P. J. Day wrote:
> On Fri, 31 Dec 2010, Maxim Levitsky wrote:
> 
> ... snip related to i915 blank screen issues ...
> 
> > Since you know how to compile kernel, and even have a close working
> > version, you really should just bisect the bug.
> >
> > Bisection is a built-in in the git binary search that allows you to
> > find the first commit that introduced the bug.
> >
> > For example, quick google gives,
> > http://www.kernel.org/pub/software/scm/git-core/docs/v1.0.13/howto/isolate-bugs-with-bisect.txt
> 
>   oh, i know how to do bisection but, to be frank, i just don't have
> the time.  i'm under a brutal deadline for some billable work and,
> even if i had the time, i suspect that when i eventually tracked down
> the commit, i probably wouldn't really appreciate what i was looking
> at, anyway.
You truly underestimate this.
Having commit that broke things points perfectly at the issue.
Then, developers can even just revert the offending commit until it is
understood why it breaks stuff.

I have done many bisections in the past, and lot of bugs were fixed that
way.

I personally can't help you with the issue, because I am nether familiar
with the hardware nor with the driver, but I just say what I do when I
get a regression (and this not uncommon).

Best regards,
	Maxim Levitsky



> 
>   i've done some searching on the issue and, among other pages, i've
> come up with this one:
> 
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/683775
> 
> so i downloaded the kernel mentioned there and tested that, but no
> luck.  mostly, i understand that there top men -- *top* *men* --
> working on this :-) who know far more about the graphics subsystem
> than i do so i'm just going to stay out of their way, but i'm willing
> to test any proposed solutions if they come up.  my biggest worry is
> that, given that we're already at -rc8, will this issue be fixed
> before the next official release?
> 
> rday
> 

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Robert P. J. Day]

On Fri, 31 Dec 2010, Maxim Levitsky wrote:

> On Fri, 2010-12-31 at 12:01 -0500, Robert P. J. Day wrote:
> > On Fri, 31 Dec 2010, Maxim Levitsky wrote:
> >
> > ... snip related to i915 blank screen issues ...
> >
> > > Since you know how to compile kernel, and even have a close working
> > > version, you really should just bisect the bug.
> > >
> > > Bisection is a built-in in the git binary search that allows you to
> > > find the first commit that introduced the bug.
> > >
> > > For example, quick google gives,
> > > http://www.kernel.org/pub/software/scm/git-core/docs/v1.0.13/howto/isolate-bugs-with-bisect.txt
> >
> >   oh, i know how to do bisection but, to be frank, i just don't have
> > the time.  i'm under a brutal deadline for some billable work and,
> > even if i had the time, i suspect that when i eventually tracked down
> > the commit, i probably wouldn't really appreciate what i was looking
> > at, anyway.

> You truly underestimate this. Having commit that broke things points
> perfectly at the issue. Then, developers can even just revert the
> offending commit until it is understood why it breaks stuff.

  one of the difficulties (if i'm reading this correctly) is that the
issue isn't even reliably reproducible.  a number of people who have
reported this claim that *sometimes* the system boots properly but
it's hit and miss, so it's not even clear you can bisect
deterministically.

rday

-- 

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: is there a fix for ubuntu with intel GPU booting to blank screen?

[Robert P. J. Day]

On Fri, 31 Dec 2010, Maxim Levitsky wrote:

> On Fri, 2010-12-31 at 12:01 -0500, Robert P. J. Day wrote:
> > On Fri, 31 Dec 2010, Maxim Levitsky wrote:
> >
> > ... snip related to i915 blank screen issues ...
> >
> > > Since you know how to compile kernel, and even have a close working
> > > version, you really should just bisect the bug.
> > >
> > > Bisection is a built-in in the git binary search that allows you to
> > > find the first commit that introduced the bug.
> > >
> > > For example, quick google gives,
> > > http://www.kernel.org/pub/software/scm/git-core/docs/v1.0.13/howto/isolate-bugs-with-bisect.txt
> >
> >   oh, i know how to do bisection but, to be frank, i just don't have
> > the time.  i'm under a brutal deadline for some billable work and,
> > even if i had the time, i suspect that when i eventually tracked down
> > the commit, i probably wouldn't really appreciate what i was looking
> > at, anyway.

> You truly underestimate this. Having commit that broke things points
> perfectly at the issue. Then, developers can even just revert the
> offending commit until it is understood why it breaks stuff.

  i suspect what i'm about to type won't come as a surprise to anyone
currently working on this problem, but there is a cheap workaround to
boot to the command line using one of the offending kernels.

  when i boot, say, 2.6.37-rc8+ (or any other kernel, for that
matter), the first two lines i always see are (paraphrasing):

/dev/pts busy ...
ureadahead info ...

and that's it, just those two lines.  it's the same no matter which
kernel i boot.  and here's where things differ.

  for a properly functioning kernel, the display then switches to
hi-res mode, the two lines are redisplayed in a smaller font and the
boot continues normally, right up to the desktop (say, with my
2.6.37-rc5+ kernel).

  for a bad kernel (say, my 2.6.37-rc8+), it's that switch to hi-res
after those original two lines of output where the display goes black
and stays that way so it looks like a KMS-related problem.

  a workaround to at least let me boot to the command line is to add

  "i915.modeset=0"

as a kernel parm, but that won't then let me start X once i'm up and
running.  there are all sorts of reports of this type of behaviour,
like this:

  https://bbs.archlinux.org/viewtopic.php?id=96759

which is where i got the idea to use that kernel parm and the
behaviour i saw matches exactly what's in that post.  in any event,
as i reported, this is still happening with -rc8.

rday

-- 

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================